For small systems we would put the zImage at 32KiB after
the start of memory, and put the DT a bit after the uImage.
The kernel will always try to relocate itself and overwrite
the DT.
Try to be more clever at uImage placement to avoid
triggering the kernel relocation.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
For small systems we would put the zImage at 8MiB after
the start of memory, and put the DT a bit after the zImage.
When we encounter an image which is bigger than 8MiB
uncompressed, the kernel would try to relocate itself
and overwrite the DT.
Try to be more clever at zImage placement to avoid
triggering the kernel relocation.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This allows to load all the lowlevel init code, including the
uncompressor, inside SRAM and not just the bare init part. This is
useful when pbl is used as a first-stage bootloader but is loaded by an
external firmware.
Signed-off-by: David Vincent <freesilicon@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This improves the initrd/devicetree placement in the bootm code.
We used to put the initrd at the start of the kernel + 8MiB. This
of course fails once the kernel gets bigger than 8MiB. Also the
place for the devicetree was allocated using malloc(). This can
lead to the problem that the devicetree is outside of the kernels
lowmem and thus not reachable for the kernel.
With this patch __do_bootm_linux gets a pointer to free space where
the devicetree and the initrd can be safely put.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
The zImage should be placed where it won't be overwritten by the
uncompressed image, otherwise the kernel decompressing code has
to relocate the zImage before decompression. As Kernels tend to
become bigger put it into 32MiB into RAM if we have enough RAM
available.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
__do_bootm_linux is called from the uImage, zImage and raw handlers.
In case of the zImage handler the kernel will already be loaded and
the kernel load code in __do_bootm_linux will do nothing. Move the
loading code to do_bootm_linux so that __do_bootm_linux will always
be called with the kernel already loaded.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
The common bootm code used to load uImage contents to SDRAM
before calling into the handlers if possible. This makes the
handlers complicated since they have to handle many cases. Instead,
introduce a helper to load the os after the handlers have figured
out a good load address.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Start a 2nd stage barebox with the Linux Kernel calling convention.
Right now barebox does not interpret ATAGs or devicetree passed
to it, but it doesn't hurt to pass parameters so that future bareboxes
can use them.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
If a board does not specify a place for the atags list default to
SDRAM start + 0x100. The vast majority of boards uses this place
anyway, so the call to armlinux_set_bootparams() can be removed
for most boards.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This adds an .oftables section right before .dtb section with
BAREBOX_CLK_TABLE to ARM linker script.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This introduces some new environment variable helpers and updates
the existing ones. Newly introduced are:
getenv_bool: read a bool variable
getenv_ul: read an unsigned long variable
getenev_uint: read an unsigned int variable
getenv_nonempty: like normal getenv, but does return NULL instead of an
empty string
All new helpers take a pointer to the value. This value is only modified
when the variable exists. This allows the following programming scheme:
unsigned int myvalue = sanedefault;
getenv_uint("myvalue", &myvalue);
So without checking the return value myvalue contains the best possible
value.
getenv_ull is updated to this scheme.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
The bootm code is now independent from the actual bootm command, so
compile the ARM specific bootm code based on CONFIG_BOOTM and not on
CONFIG_CMD_BOOTM.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This patch placed the flattened devicetree to armlinux_bootparams.
armlinux_bootparams normally is at SDRAM_START + 0x100. The kernels
initial page tables are normally at SDRAM_START + 0x4000, so the
flattened devicetree gets overwritten once it exceeds 0x3f00 bytes
which is quite common.
Revert this patch for now once a better solution can be found
This reverts commit 0c4108f917.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Alexander Shiyan <shc_work@mail.ru>
The Kernel commandline is an important debugging aid when people
ask why their Kernel won't start, so print it unconditionally.
This is done in !dt mode anyway, so also do it with dt.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
In some cases, the address of the devicetree, which is obtained
dynamically, may be located in the kernel text area, which leads
to overwrite devicetree by kernel. The patch uses the address for
devicetree, provided by board, if it possible.
Signed-off-by: Alexander Shiyan <shc_work@mail.ru>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
In the upcoming multi image build process we will cat images together.
To find the concatenated image we need to reliably find the end of the
current binary. This adds a dummy section at the end of a pbl binary.
Its only purpose is to mark the end of the image. The multi image
patches will add something to this section so that it doesn't get
discarded by the linker.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Since the pbl linker script can be reused by the upcoming multi image
build process move it to a common place. Also remove ENTRY() from the
linker script and instead add the -e option to the linker. This makes
the entrypoint configurable.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Without devicetree support we print the Kernel commandline in
verbose mode. Do the same with devicetree boot aswell.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
when doing bootm -v -v we dumped the original tree to the console.
Make sure to print the fixed tree instead so that the fixups can
be examined.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
From Kernel commit 418df63a ARM: 7670/1: fix the memset fix
| Commit 455bd4c430b0 ("ARM: 7668/1: fix memset-related crashes caused by
| recent GCC (4.7.2) optimizations") attempted to fix a compliance issue
| with the memset return value. However the memset itself became broken
| by that patch for misaligned pointers.
|
| This fixes the above by branching over the entry code from the
| misaligned fixup code to avoid reloading the original pointer.
|
| Also, because the function entry alignment is wrong in the Thumb mode
| compilation, that fixup code is moved to the end.
|
| While at it, the entry instructions are slightly reworked to help dual
| issue pipelines.
|
| Signed-off-by: Nicolas Pitre <nico@linaro.org>
| Tested-by: Alexander Holler <holler@ahsoftware.de>
| Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
- Add rules to generate dtb files in arch/arm/dts/
- add an initcall which unflattens and probes the internal devicetree
- Add skeleton devicetree
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
For making the same binary executable on different SoCs which have
different DRAM addresses we have to be independent of the compile
time link address.
This patch adds relocatable binary support for the ARM architecture.
With this two new functions are available. relocate_to_current_adr
will fixup the binary to continue executing from the current position.
relocate_to_adr will copy the binary to a given address, fixup the
binary and continue executing from there.
For the PBL and the real image relocatable support can be enabled
independently. This is done to (hopefully) better cope with setups
where the PBL runs from SRAM or ROM and the real binary does not.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
With relocatable binaries the linker is not able to supply absolute
addresses. These only get available when the relocation function is
being run. Since for early initialization we need some variables
before relocation, we supply them relatively to some known address
in the binary. This means that the variables have to be converted
to absolute addresses during runtime.
This patch adds a C macro and an assembly macro to calculate the
variables during runtime.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This makes it possible to modify the tree in the handlers.
This is necessary because the initrd addresses are only
known inside the handlers, but not to the generic bootm
code.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Without compiled in devicetree support we used to copy the
concatenated devicetree directly behind the zImage. This is
unnecessary, even if we do not have devicetree support we can
copy the devicetree whereever we like and pass the kernel a
pointer to it. This makes the code a bit easier.
While at it, add the missing free calls in the error case.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
With this every devicetree is first converted to the barebox internal
format before it's converted back to dtb again.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Recent GCC versions (e.g. GCC-4.7.2) perform optimizations based on
assumptions about the implementation of memset and similar functions.
The current ARM optimized memset code does not return the value of
its first argument, as is usually expected from standard implementations.
For instance in the following function:
void debug_mutex_lock_common(struct mutex *lock, struct mutex_waiter *waiter)
{
memset(waiter, MUTEX_DEBUG_INIT, sizeof(*waiter));
waiter->magic = waiter;
INIT_LIST_HEAD(&waiter->list);
}
compiled as:
800554d0 <debug_mutex_lock_common>:
800554d0: e92d4008 push {r3, lr}
800554d4: e1a00001 mov r0, r1
800554d8: e3a02010 mov r2, #16 ; 0x10
800554dc: e3a01011 mov r1, #17 ; 0x11
800554e0: eb04426e bl 80165ea0 <memset>
800554e4: e1a03000 mov r3, r0
800554e8: e583000c str r0, [r3, #12]
800554ec: e5830000 str r0, [r3]
800554f0: e5830004 str r0, [r3, #4]
800554f4: e8bd8008 pop {r3, pc}
GCC assumes memset returns the value of pointer 'waiter' in register r0; causing
register/memory corruptions.
This patch fixes the return value of the assembly version of memset.
Could you please review, or suggest better alternatives ?
Thanks,
--
Ivan
(this is a shorter and (hopefully) clearer repost of
http://lists.infradead.org/pipermail/linux-arm-kernel/2013-January/144916.html)
The patch adds a 'mov' instruction and merges an additional load+store into
existing load/store instructions.
For ease of review, here is a breakdown of the patch into 4 simple steps:
Step 1
======
Perform the following substitutions:
ip -> r8, then
r0 -> ip,
and insert 'mov ip, r0' as the first statement of the function.
At this point, we have a memset() implementation returning the proper result,
but corrupting r8 on some paths (the ones that were using ip).
Step 2
======
Make sure r8 is saved and restored when (! CALGN(1)+0) == 1:
save r8:
- str lr, [sp, #-4]!
+ stmfd sp!, {r8, lr}
and restore r8 on both exit paths:
- ldmeqfd sp!, {pc} @ Now <64 bytes to go.
+ ldmeqfd sp!, {r8, pc} @ Now <64 bytes to go.
(...)
tst r2, #16
stmneia ip!, {r1, r3, r8, lr}
- ldr lr, [sp], #4
+ ldmfd sp!, {r8, lr}
Step 3
======
Make sure r8 is saved and restored when (! CALGN(1)+0) == 0:
save r8:
- stmfd sp!, {r4-r7, lr}
+ stmfd sp!, {r4-r8, lr}
and restore r8 on both exit paths:
bgt 3b
- ldmeqfd sp!, {r4-r7, pc}
+ ldmeqfd sp!, {r4-r8, pc}
(...)
tst r2, #16
stmneia ip!, {r4-r7}
- ldmfd sp!, {r4-r7, lr}
+ ldmfd sp!, {r4-r8, lr}
Step 4
======
Rewrite register list "r4-r7, r8" as "r4-r8".
Signed-off-by: Ivan Djelic <ivan.djelic@parrot.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Add the initrd start and end address to the DT, code comes from u-boot.
Signed-off-by: Vicente Bergas <vicencb@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Hi Sascha,
I've made the changes you suggested in this resent patch.
Everything related to custom ATAGs has been moved to the board
directory.
The generic code does not make any references to feature lists or
bootloader versions.
About the setup_feature_list prototype:
it has been renamed to atag_appender
it's not a function, it's a pointer to a function. Can it have a
prototype other than it's own declaration?
All non-related changes has been dropped. They were checkpatch.pl
warnings unrelated to this patch.
Regards,
Vicente.
Signed-off-by: Vicente Bergas <vicencb@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
When an oftree is already specified use it. This lets the user
boot a kernel with an oftree he provided himself rather than
hardcoding the concatenated one.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Add barebox-data section in arm branch to get complete
barebox regions in sdram regions tree.
Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
With bootm of_fix_tree() will already be called from the generic bootm
code, so do not do this again in the Android image handler.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Currently the bootm code uses of_fix_tree to apply the fixups
to the devicetree given on the command line. This function assumes
that there is enough space for the fixups available. Also on ARM
we have to make sure the tree does not cross 1Mib boundaries.
This patch moves the space allocation and alignment ensurance
to of_get_fixed_tree and uses it in bootm. This is the first
step for making of_get_fixed_tree the single point of devicetree
handling in barebox.
of_get_fixed_tree now takes an argument of the input fdt. If it is
given, this one is used, otherwise an internal oftree is used which
will be created in subsequent patches.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Pass the buffer size to the file detection code. This makes sure we do not
read past the buffer. This is especially useful for ext filesystem detection
as the magic is at byte offset 1080. Also introduce a FILE_TYPE_SAFE_BUFSIZE
define which is set to the minimum bufsize the detection code needs to detect
all known filetypes.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Lucas Stach
David Vincent
Sascha Hauer
Sebastian Hesselbarth
Alexander Shiyan
Nicolas Pitre
Ivan Djelic
Vicente Bergas
Alexander Aring